Economic evaluations of hepatitis B vaccination strategies - A systematic review of the literature | RIVM

Hele tekst

(1)UHVHDUFKIRU PDQDQGHQYLURQPHQW. RIJKSINSTITUUT VOOR VOLKSGEZONDHEID EN MILIEU NATIONAL INSTITUTE OF PUBLIC HEALTH AND THE ENVIRONMENT. RIVM report 403505 003 (FRQRPLFHYDOXDWLRQRIKHSDWLWLV%YDFFLQDWLRQ VWUDWHJLHV. A systematic review of the literature G.A. de Wit, R. Welte November 1999. This investigation has been performed by order and for the account of the ministry of Health, Welfare and Sports, within the framework of project 403505, Cost-effectiveness of interventions in care and prevention. RIVM, P.O. Box 1, 3720 BA Bilthoven, telephone: 31 - 30 - 274 91 11; telefax: 31 - 30 - 274 29 71.

(2) page 2 of 79. RIVM report 403505 003.

(3) RIVM report 403505 003. page 3 of 77. $EVWUDFW National and international economic evaluations of universal vaccination against hepatitis B were systematically selected from the literature. Only those studies were selected that had (a) reported original data in Dutch, English, French or German; (b) evaluated at least one universal vaccination strategy; (c) shown a high methodological quality and (d) been conducted in countries with a hepatitis B epidemiological pattern comparable to the Dutch situation. Sixteen relevant studies were identified out of 1060 publications registered in the databases COCHRANE LIBRARY, CURRENT CONTENTS, DARE, HEED, INAHTA DATABASE, MEDLINE, NEED and by applying the system of reference tracking. Despite the strict inclusion criteria, strong divergence is shown with regard to study assumptions, methodology, outcomes and conclusions. Some studies show universal screening to be cost-saving, while others report unfavourable cost-effectiveness ratios. Studies that investigated similar vaccination strategies reach dissimilar conclusions on the most cost-effective strategy. Several parameters were found to usually have a large impact on the results and conclusions of a study. These are: a) vaccine costs, b) discount rate, c) hepatitis B incidence, d) inclusion of indirect costs and effects, e) percentage of acute infections that lead to chronic hepatitis, f) discounting of effects and h) assumed lifelong costs of hepatitis B infection, including long-term consequences such as cirrhosis and liver cancer. Hence, in any further study these important variables should be subject to extensive sensitivity analysis. Economic evaluations that are intended to assist in local policy-making should take careful account of local circumstances. The 16 selected studies have little practical relevance for Dutch policy makers..


(5) RIVM report 403505 003. page 5 of 77. 3UHIDFH This study on the economic evaluation of hepatitis B vaccination strategies forms part of the RIVM project concerning the ‘efficiency of interventions in prevention and care’ and was performed at the department for Health Services Research CZO. The study is related to other RIVM projects on the epidemiology and the mathematical modelling of the transmission of infectious diseases performed at the department for Infectious Diseases Epidemiology CIE. The work on hepatitis B has been performed by order of the Ministry of Health, Welfare and Sports / Inspectorate for Health Care. With respect to the cost-effectiveness of interventions, CZO co-operates with the Institute for Medical Technology Assessment of the Erasmus University Rotterdam and this report is produced within the context of this co-operation. The study also forms part of the project on the ‘impact and costs of hepatitis B/C and HIV in intravenous drug users’ (CT.98.EP.06), funded by the European Monitoring Centre for Drugs and Drug Addiction (EMCDDA). Dr. J.C. Jager, projectleader.


(7) RIVM report 403505 003. page 7 of 77. &RQWHQWV 6DPHQYDWWLQJ. . 6XPPDU\. . . ,QWURGXFWLRQ. . . %DFNJURXQG. . . . +HSDWLWLV%YLUXV. . . (SLGHPLRORJ\. . . 3UHYHQWLRQ. . 'HVLJQDQGFRQGXFWRIOLWHUDWXUHUHYLHZ. . $LPRIWKHOLWHUDWXUHUHYLHZ. . . 'HVLJQRIWKHOLWHUDWXUHUHYLHZ. . 3.2.1 3.2.2 3.2.3 . Criteria for selection of papers Databases used to identify relevant publications Further procedures. 5HVXOWV. 22 23 24 . . 5HVXOWVRIVHOHFWLRQSURFHVV. . 0DLQFKDUDFWHULVWLFVRIVHOHFWHGSDSHUV. 4.2.1 4.2.2 4.2.3 4.2.4 4.2.5 4.2.6 4.2.7 4.2.8 4.2.9 4.2.10 4.2.11 . . Vaccination strategy Country of study and its epidemiological pattern of HBV spread Variables considering the vaccination program Type of economic evaluation and perspective of study Modelling techniques Duration of vaccination programs and time horizon for calculation of costs and effects Costs of vaccine Different cost data included in studies Discounting Sensitivity analysis Conclusions reached on economic attractiveness of evaluated programs. *HQHUDOREVHUYDWLRQVDQGGLVFXVVLRQ. . . 26 28 29 31 33 33 35 36 37 38 41 . 5HIHUHQFHV. . /LVWRIDEEUHYLDWLRQV. . $SSHQGL[. . 0DLOLQJOLVW.

(8) page 8 of 79. RIVM report 403505 003. $SSHQGL[. &KHFNOLVWWRHYDOXDWHHFRQRPLFVWXGLHV. . $SSHQGL[. 3DUDPHWHUHVWLPDWHIRUP. . $SSHQGL[. 'HVFULSWLRQRIVWXG\FKDUDFWHULVWLFV. . $SSHQGL[. 6KRUWUHYLHZVRIVHOHFWHGVWXGLHV LQFKURQRORJLFDORUGHU

(9). . $SSHQGL[. ([FOXGHG6WXGLHV LQDOSKDEHWLFDORUGHU

(10). .

(11) RIVM report 403505 003. page 9 of 77. 6DPHQYDWWLQJ Hepatitis is een leverontsteking die onder andere veroorzaakt kan worden door een virusinfectie, alcohol of drugs. Dit rapport gaat over de preventie van hepatitis die veroorzaakt wordt door het hepatitis B virus (HBV). Sinds 1986 is een vaccin beschikbaar dat geproduceerd wordt door middel van recombinant DNA technieken. Dit vaccin maakt bescherming tegen infectie mogelijk bij personen die nog niet eerder met het virus in aanmerking kwamen, en daarmee ook bescherming tegen de lange-termijn gevolgen van besmetting met het virus, zoals levercirrose en leverkanker. Het Hepatitis B virus kent heel verschillende verspreidingsvormen in verschillende regio’s van de wereld. Nederland behoort, samen met andere WestEuropese landen en Noord-Amerika, tot de laag-endemische gebieden. Rond de 5 procent van de bevolking komt tijdens het leven met HBV in aanmerking. Risico-groepen zijn onder anderen intraveneuze druggebruikers, dialysepatienten, homoseksuelen en personen met veel wisselende sexuele contacten. Zwangeren die drager zijn van het virus kunnen het virus tijdens of na de geboorte overgeven aan hun kind. Daarom worden zwangere vrouwen in Nederland sinds 1989 gescreend op aanwezigheid van het virus. Indien zij drager zijn worden hun kinderen onmiddellijk na de geboorte passief en actief geïmmuniseerd. De Wereldgezondheidszorgorganisatie WHO beveelt sinds 1992 aan dat alle landen pasgeborenen vaccineren met het Hepatitis B vaccin. Dit is weliswaar kostbaar, maar op korte en lange termijn zouden daarmee ook kosten bespaard kunnen worden omdat minder mensen geïnfecteerd worden en ziekteverschijnselen ontwikkelen. Bovendien kan de ziektelast van Hepatitis B, in termen van voortijdige sterfte en verloren levensjaren, aanmerkelijk gereduceerd worden. In veel landen werden reeds economische evaluatiestudies uitgevoerd, ter ondersteuning van de besluitvorming rondom de invoer van dergelijke algemene vaccinatiecampagnes. Deze economische studies zijn gericht op het in kaart brengen van de kosten en effecten van vaccinatie op korte en (middel-)lange termijn. In dit rapport wordt een overzicht gegeven van het “study design” en de uitkomsten van 16 gepubliceerde studies, die als onderwerp de economische aspecten van een of meerdere algehele vaccinatiestrategieën hadden. Deze studies zijn zorgvuldig geselecteerd uit een veel groter arsenaal van studies die handelen over economische aspecten van hepatitis B. Zeven verschillende databases werden geraadpleegd om relevante studies te identificeren: COCHRANE LIBERARY, CURRENT CONTENTS, DARE, HEED, INAHTA DATABASE, MEDLINE, NEED. Verder werd het systeem van “reference tracking” (sneeuwbalmethode) toegepast om nog niet eerder gevonden publicaties op te sporen. In drie tijdschriften die niet in bovenstaande databases zijn opgenomen en waarvan verwacht werd dat ze relevante bijdragen zouden kunnen bevatten werd handmatig gezocht. De volgende criteria werden gehanteerd voor de selectie van studies:.

(12) page 10 of 79. •. • • •. •. RIVM report 403505 003. oorspronkelijk werk over economische aspecten van hepatitis B preventie, waarin minimaal een algemene vaccinatiestrategie wordt geëvalueerd. De vaccinatiestrategie moet gericht zijn op een specifieke doelgroep, een regio of een land; Taal: Duits / Engels / Frans / Nederlands; Beschikbaar via Nederlandse bibliotheken; De studie moet uitgevoerd zijn in een land waar de epidemiologie van hepatitis B niet al te zeer afwijkt van de Nederlandse epidemiologische situatie. Een serum-prevalentie van 8 procent werd als bovengrens gehanteerd; De studie moet voldoen aan algemeen geaccepteerde methodologische richtlijnen voor economische evaluaties. De methodologische kwaliteit werd gekwantificeerd door middel van een lijst van 13 criteria, waarvoor 0 tot 4 punten verdiend konden worden. Studies waarvan het gemiddelde puntenaantal per van toepassing zijnd item 3 of hoger was werden geselecteerd voor deze literatuur review.. De diverse zoekstrategieën leverden 1060 referenties op. Daarvan kon een groot deel op grond van de opgestelde in-en uitsluitingscriteria onmiddellijk als niet relevant voor deze literatuurstudie worden aangemerkt. Een aantal van 84 artikelen en rapporten is wel gedegen bestudeerd. Een groot deel daarvan kon op basis van de kwaliteitscriteria terzijde worden gelegd. Uiteindelijk werden 16 studies geselecteerd voor dit literatuuroverzicht. Deze studies werden gepubliceerd tussen 1982 en 1998. Vijftien van de zestien studies werden echter in de jaren ’90 gepubliceerd. Van deze studies worden de volgende elementen systematisch beschreven in dit rapport: • de onderzochte vaccinatiestrategie • de doelgroep van het vaccinatieprogramma • het land en regio waarin vaccinatieprogramma wordt uitgevoerd • de incidentie van HBV in dat land / regio • de serumprevalentie van hepatitis markers in dat land / regio • diverse veronderstellingen omtrent de effectiviteit van het vaccinatieprogramma: • de duur van de immuniteit na vaccinatie • de werkzaamheid van het vaccin • “compliance” met het vaccinatieprogramma • het percentage van de doelgroep dat succesvol wordt gevaccineerd • eventuele bijwerkingen van het vaccin • het type economische evaluatie • het perspectief van de studie • de gebruikte modelleringstechniek(en) • de duur van het vaccinatieprogramma • de tijdshorizon voor het meten van effecten van het vaccinatieprogramma • de tijdshorizon voor het meten van kosten en baten van het vaccinatieprogramma.

(13) RIVM report 403505 003. • • • • • • • • • • • •. page 11 of 77. de veronderstelde kosten van het vaccin de bron die gebruikt wordt voor de kosten van het vaccin het basisjaar voor de kostencalculatie de munteenheid die gehanteerd wordt welke kostencategorieën worden meegenomen in de analyse de gehanteerde discontvoet het al dan niet disconteren van effecten welke variabelen worden onderzocht in een sensitiviteitsanalyse de variabelen die in de sensitiviteitsanalyse een belangrijke invloed op de resultaten van de studie blijken te hebben de gerapporteerde kosten-effectiviteits-ratio’s de gerapporteerde baten-kosten-ratio’s de conclusies die worden getrokken. Ondanks de toepassing van strikte selectiecriteria blijken de 16 geselecteerde studies sterk te verschillen qua onderzoeksmethodologie en ook qua rapportage. Sommige studies geven in uitvoerige bijlagen een zorgvuldige onderbouwing voor de keus van de parameterwaarden, andere studies refereren simpelweg aan gemaakte veronderstellingen in eerder gepubliceerde studies. De grootste verschillen tussen studies werden gevonden op de volgende terreinen: • de veronderstelde duur van de immuniteit (tussen 3 jaar en levenslang) • het percentage van de doelgroep dat bereikt wordt (tussen 26 % en 100 %) • de gebruikte modelleringtechniek (eenvoudige berekeningen van vermeden infecties versus combinaties van Markov-keten-analyse en Monte Carlo simulatie met meenemen van “herd immunity” effecten) • de duur van de interventie (van 1 tot 100 jaar) • de tijdshorizon voor de berekening van effecten (van 1 tot 100 jaar) • de tijdshorizon voor de berekening van kosten (van 1 tot 129 jaar) • de veronderstelde kosten van het vaccin (US$ 6 tot US$ 225) • het al dan niet includeren van indirecte kosten (productiviteitskosten) • het al dan niet disconteren van effecten van vaccinatie • de gebruikte discontovoet (van 0 tot 8 %) Vanwege deze enorme verschillen is het moeilijk om de in de diverse studies getrokken conclusies rechtstreeks met elkaar te vergelijken. De conclusies zijn dan ook heel verschillend: sommige studies concluderen dat algemene vaccinatiecampagnes kostenbesparend zouden kunnen worden georganiseerd, andere studies rapporteren juist extreem hoge kosteneffectiviteits-ratio’s. Studies waarin meerdere vaccinatiestrategieën met elkaar worden vergeleken trekken verschillende conclusies omtrent de meest kosten-effectieve strategie: soms is dat de algehele vaccinatie van pasgeborenen, soms een op adolescenten gerichte strategie en soms combinaties van beide strategieën. De twee studies waarin vaccinatiecampagnes gericht op mannelijke homoseksuelen en intraveneuze druggebruikers worden geëvalueerd conclude-.

(14) page 12 of 79. RIVM report 403505 003. ren dat het waarschijnlijk mogelijk is om dit kosten-neutraal of zelfs kostenbesparend te doen. Vanwege de enorme verschillen tussen studies is het in ieder geval niet mogelijk om conclusies die getrokken worden in een bepaalde setting rechtstreeks naar een andere setting te vertalen. Elke beleidsbeslissing die genomen moet worden omtrent het al dan niet invoeren van algemene vaccinatie tegen hepatitis B vereist een zorgvuldige studie waarin de locale omstandigheden zo gedetailleerd mogelijk worden meegenomen. De geselecteerde studies hebben in het algemeen dan ook weinig praktische relevantie voor in Nederland te nemen beleidsbeslissingen. Omdat de studies zo sterk verschillen is het belangrijk dat de invloed van de keuze van parameterwaarden op het eindresultaat van de studie zichtbaar gemaakt wordt. Alle studies besteden hieraan aandacht in min of meer uitgebreide sensitiviteitsanalyses. De volgende parameters worden het meest genoemd als belangrijke beïnvloeders van de uitkomsten van de studies: 1. 2. 3. 4. 5. 6. 7.. De kosten van het vaccin De gehanteerde discontovoet De incidentie van hepatitis B Het al dan niet meenemen van indirecte kosten/effecten Het percentage infecties dat resulteert in een chronische besmetting met HBV Het al dan niet disconteren van effecten De veronderstelde levenslange kosten van hepatitis B besmetting, inclusief lange termijn gevolgen zoals levercirrose en leverkanker. Het verdient aanbeveling dat in toekomstige economische evaluaties in elk geval de invloed van bovenstaande variabelen gedetailleerd wordt onderzocht..

(15) RIVM report 403505 003. page 13 of 77. 6XPPDU\ Hepatitis is an inflammation of the liver that may be caused by a variety of agents including viruses, alcohol and drugs. This report focuses on the prevention of hepatitis that is generated by the hepatitis B virus (HBV). A vaccine against HBV that is produced using recombinant DNA technology is available since 1986. It enables the protection of persons that have not yet been infected by the virus. It also prevents long term complications of HBV-infection such as cirrhosis and liver cancer. The predominant modes of HBV transmission vary throughout the world. With other European countries and the US, the Netherlands belongs to countries with low HBV endemicity. Some 5% of the inhabitants of the Netherlands will become infected with HBV during their lifetime. Intravenous drug users, dialysis-patients, homosexuals, and people with many different sex-partners are at increased risk for HBV infection. Infected pregnant women can transfer the virus during birth (vertical or perinatal transmission). Because of that, pregnant women are screened for HBV in the Netherlands since 1989. If they are HBV-carrier their offspring will receive active and passive immunisation. Since 1992 the World Health Organisation recommends that worldwide all newborns should be vaccinated with the hepatitis B vaccine. This is expensive, but sooner or later medical expenditures will be prevented, as fewer persons are infected and develop disease symptoms. Furthermore, the burden of hepatitis B, in terms of premature deaths and loss of life-years can be reduced significantly. In many countries, economic evaluation studies have been conducted to support decision making about the implementation of such vaccination programmes. These economic studies are aimed at the investigation of costs and effects of vaccination on the short, (middle) and long term. In this report we present an overview of the study design and results of 16 published studies that examine the economic aspects of one or several universal vaccination strategies. These studies have carefully been selected from a much bigger pool of studies that are concerned with the economic aspects of hepatitis B. Seven different databases have been searched to identify the relevant studies: COCHRANE LIBRARY, CURRENT CONTENTS, DARE, HEED, INAHTA DATABASE, MEDLINE, NEED. In addition, the system of reference tracking was used to find publications that were not covered in these databases. Finally, the indexes of three journals that were expected to contain relevant papers and were known not to be indexed in the databases used were searched manually. The following criteria have been used to select the studies: • Original study about the economic aspects of hepatitis B prevention that evaluates minimally one universal vaccination strategy. The vaccination strategy has to be aimed at a specific target population, a region or a country; • Language: Dutch, English, French, German; • Available from Dutch libraries;.

(16) page 14 of 79. •. •. RIVM report 403505 003. The study has to be conducted in a country in which the epidemiology of hepatitis B differs not too much from the Dutch epidemiological situation. A serum prevalence of 8 percent has been used as upper limit for this criterion; The study has to follow commonly accepted methodological guidelines for economic evaluations. The methodological quality is quantified by employing a list of 13 criteria. For each criterion points were given, from 0 (low quality) to 4 (high quality). Only studies that reached an average score of 3.0 or higher were selected for this literature review.. The various search strategies resulted in 1060 references. The majority of these studies could be identified as not relevant by applying the above described inclusion / exclusion criteria. Most of the remaining 84 studies did not meet the high methodological standards required. Finally, 16 studies were selected for the literature review. These studies have been published between 1982 and 1999. Fifteen of the sixteen studies are even published in the 1990s. The following elements of these studies are systematically described in the report: • Investigated vaccination strategy • Target population of the vaccination programme • Country and region of study • Incidence of HBV in that country or region • Serum-prevalence of hepatitis B markers in that country or region • Some assumptions about the effectiveness of the vaccination programme: • Duration of immunity after vaccination • Vaccine-efficacy • Compliance with the vaccination programme • Percentage of the target population that is successfully vaccinated • Possible side-effects of the vaccine • Type of economic evaluation • Study-perspective • Applied modelling technique(s) • Duration of the vaccination programme • Time-horizon for measuring effects of the vaccination programme • Time-horizon for measuring costs and benefits of the vaccination programme • Assumed costs of the vaccine • Sources of the vaccine cost data • Basic year for the cost-calculation • Currency unit • Included cost categories • Discount rate • Discounting of effects • Variables investigated in the sensitivity analysis.

(17) RIVM report 403505 003. • • • •. page 15 of 77. Variables that were identified in the sensitivity analysis to influence the results significantly Presented cost-effectiveness ratios Presented cost-benefit ratios Conclusions. Despite the application of strict selection criteria, the 16 selected studies diverge with respect to their methodology and reporting. Some studies supply with comprehensive information and a solid basis for the chosen parameter values. Other studies simply refer to the assumptions that have been made in the literature. The biggest differences between the 16 studies have been observed in the following topics: • Assumptions on the duration of the immunisation (between 3 years and life-long immunity) • The percentage of the target population that can be reached (between 26 and 100%) • Applied technique for modelling (simple calculation of averted infections versus combinations of Markov-chain-analysis and Monte Carlo simulation considering herd immunity effects) • Duration of the intervention (1 to 100 years) • Time horizon for the cost calculation (1 to 129 years) • Assumed vaccine costs (US$ 6 to US$ 225) • whether or not indirect costs (productivity costs) were included • whether or not effects were discounted • Discount rate (0 – 8%) Because of these big differences it is difficult to compare the conclusions of the selected studies. Also, conclusions reached are sometimes contradictory: some studies point out that universal vaccination may be cost saving while others report unfavourably high costeffectiveness ratios. Studies that investigate several vaccination strategies reach dissimilar conclusions regarding the most cost-effective strategy. Some studies suggest universal vaccination of new-borns is most cost-effective. Others show that vaccination of adolescents has the lowest cost-effectiveness ratio. Finally, there are also studies that favour a combination of both strategies. Two studies were conducted to evaluate HBV-vaccination of homosexual men and intravenous drug users. They reach the conclusion that these programmes are likely to be cost-neutral or even cost saving. Due to the enormous differences between the studies it is not easy to apply conclusions that have been reached in one particular setting also in another setting. Any policy-decision about the introduction of universal HBV screening requires a careful analysis. The local circumstances should be taken into account as close as possible. Thus, the selected studies generally have little relevance for policy making in the Netherlands..

(18) page 16 of 79. RIVM report 403505 003. As the selected studies differ so much it is important to show the influence of parameter values on the study results. For this, all studies employ sensitivity analyses. The following parameters have most often been identified as sensitive variables: 1. Vaccine costs 2. Discount rate 3. Hepatitis B incidence 4. Exclusion of indirect costs / effects 5. Percentage of infections that lead to chronic HBV infections 6. Not discounting of effects 7. Assumed lifelong costs of HBV infection, including long-term consequences such as cirrhosis and liver cancer It should be noted that in future economic evaluations the influence of the factors above should always be investigated in detail..

(19) RIVM report 403505 003. page 17 of 77. ,QWURGXFWLRQ. This report describes the results of a literature review on the economic attractiveness of general vaccination programs directed at limiting the burden of hepatitis B to society. In Chapter 2, the reader will be introduced to the disease and its causes, the epidemiological patterns as found in various regions of the world and in the Netherlands in particular, and to several strategies to prevent hepatitis B. Chapter 3 contains a description of the aim and conduct of the literature review. In- and exclusion criteria, the search strategy and databases that were consulted are described. Chapter 4 contains the results of the study. Firstly, the result of the selection process is explained. Secondly, the main characteristics of the 16 selected studies are described in a systematic way. In Chapter 5, general observations and a discussion of the results of the study can be found. Several Appendices provide with background material to the study..

(20) page 18 of 79. RIVM report 403505 003. %DFNJURXQG. +HSDWLWLV%YLUXV Hepatitis is a general term for an inflammation of the liver that may be caused by viruses, bacteria, drugs, toxins, metabolic disease or excess alcohol intake. Individuals with hepatitis tend to have generalised symptoms that resemble symptoms of the early stages of flu, such as fatigue, nausea, fever, diarrhoea and jaundice. As the disease progresses, other symptoms may occur: chills, weight loss, distaste for food, extreme fatigue, pain in the right upper quadrant and dark urine. It is only since 10 years that the 5 different viruses (hepatitis A, B, C, D and E) that are currently known to cause viral hepatitis may be distinguished. Hepatitis A and B have been known since the 1940s. The other three viruses have been identified in the 1970s and late 1980s. Because symptoms are not specific to the causative agent, it is impossible to base a diagnosis on clinical symptoms alone. Serological testing is required. This report is only concerned with hepatitis B (formerly “serum hepatitis”), the hepatitis virus that has most serious health consequences. HBV is transmitted by infected blood and other body fluids, e.g. semen and vaginal secretions. Perinatal, horizontal (e.g. household-contacts), sexual, and parenteral (e.g. skin penetration with contaminated medical, dental, tattooing or piercing devices) transmission ways are currently recognised. Risk groups include injecting drug users, persons with multiple sex partners or a history of sexually transmitted diseases, health care workers who have occupational contact with blood, haemodialysis patients, infants born to HBV infected mothers, institutionalised populations, recipients of unscreened blood or blood-derived products and household contacts of HBV infected individuals. To identify HBV and to determine the status of an HBV-infection, serologic markers of HBV are used. Among them are hepatitis B surface antigen (HBsAg), hepatitis B core antigen (HBcAg), hepatitis B e antigen (HBeAg) and the corresponding antibodies. The laboratory criteria for diagnosis of an acute HBV infection are 1) presence of HBsAg or IgM antibody to HBcAg, and 2) absence of IgM anti-hepatitis A antigen. HBsAg indicates active viral replication and shows that a person is infectious. HBsAg presence for 6 months or a positive test for HBsAg with a negative test for IgM anti-HBc indicates that the chronic carrier state has developed. HBeAg is positively associated with the risk of transmission. Mothers who are positive for HBsAg and HBeAg have a risk of 0.7 to 0.9 to infect their newborn baby perinatally [Beasly, 1981]. If a mother is positive for HBsAg but negative for HBeAg, the risk of perinatal transmission is reduced to 0.2 or less..

(21) RIVM report 403505 003. page 19 of 77. HBV causes acute and chronic hepatitis. The chances of becoming chronically infected depend upon age. About 90 percent of infected neonates and 30-50 percent of infected small children will become chronically infected. [Edmunds, 1993] In contrast, 10 to 15 % of adolescents and 2 to 8 % of adults infected with HBV develop chronic hepatitis B. In some individuals who become chronically infected, especially neonates and children, the acute infection will not be clinically apparent. Acute hepatitis can range from subclinical disease to fulminate hepatic failure. Patients with chronic HBV are potentially infectious, may have clinically apparent chronic hepatitis and are at risk for developing cirrhosis (necrosis of liver cells) and primary liver cancer (hepatocellular carcinoma or HCC).. (SLGHPLRORJ\ The hepatitis B virus is carried by an estimated 2 to 3 billion people in the world, and is the major cause of liver cancer. [Blumberg, 1975] Hepatitis B ranks second only to tobacco smoking as a cause of global cancer mortality. Hepatitis B is characterised by different epidemiological patterns in different regions of the world. Overall, incidence is lowest in Western Europe and other developed countries and highest in Asian countries. Within Europe, 3 different types of hepatitis B epidemiological patterns may be distinguished, based on different chronic HBsAg carriage rates. [Goudeau, 1990] In general, the level of endemicity increases from north to south and from west to east. The 3 different types are shown in Table 1..

(22) page 20 of 79. RIVM report 403505 003. Table 1: Epidemiological patterns of hepatitis B in Europe Characteristics. Country Type 1 <5%. Type 2 5 - 10 %. Type 3 10 - 20 %. < 0.1 %. 0.1 – 5 %. Main transmission route. None. Sexual. North Europe: 0.5 % South Europe: 5 % Perinatal or at young age. Transmission from HBsAg + mothers. None. ± 1 % of births. Overall prevalence of markers Prevalence of HBsAg. ± 3 % of births. Source: Goudeau, 1990. In typical type 1 countries, incidence of hepatitis B has always been low and is further declining as a result of successful immunisation of high-risk groups, such as health care workers and intravenous drug users. In type 2 countries, sexual transmission is the primary source of infection. Some type 2 countries have been able to reduce the perinatal transmission by means of screening of pregnant women. In type 3 countries, incidence occurs mostly before the age of 5 years, during birth or within families (household contacts). The Netherlands may be characterised mainly as a type 2 country, although the introduction of prenatal screening (in 1989) has reduced the percentage of new-borns with HBV infection significantly. In 1989-1990, hepatitis B prevalence (HBsAg +) among pregnant women was 0.45 %. [Grosheide, 1991] Recent research showed that 2.1 % of the Dutch population had markers of current of past HBV infection and that the serum-prevalence of HBsAg was 0.2 %. [van Marrewijk, 1999] Annual incidence of acute hepatitis B, as reported to the Inspectie voor de Gezondheidszorg (Inspectorate for Health Care) is about 1.45 cases per 100.000 of the population. [Grosheide, 1991] The incidence among males is about 2.3 times higher than among females. Incidence occurs most in persons between 15 and 44 years of age. However, these incidence data are unreliable due to inadequate diagnosis and reporting of hepatitis B infection.. 3UHYHQWLRQ Spread of the infection may be prevented through the screening of blood and blood products, adequate sterilisation of medical equipment, destruction of disposables, effective use of barrier techniques (condoms, gloves) and education about the disease and its risk factors. Temporary passive prophylaxis is available through the Hepatitis B Immune Globuline (HBIG). Its immediate use is recommended for infants born to HBV infected mothers, sex partners of individuals with acute hepatitis B, and in unvaccinated health care workers who experienced a needle stick injury. Post-exposure prophylaxis with HBIG is advocated along with the hepa-.

(23) RIVM report 403505 003. page 21 of 77. titis B vaccine, which offers active and prolonged immunity. The first hepatitis B vaccine available (1982) was produced from the plasma of persons chronically infected with HBV. The vaccines that are being used since 1985 are produced by recombinant DNA technology. The conventional regimen is a series of three intramuscular doses given over a six-month period: initial vaccination, again at 30 days and at 6 months. This regimen induces antibody re-sponse (measured as the persistence of anti-HBs in serum) in 85-90 percent of adults and in over 95 percent of infants and children. The persistence of anti-HBs in serum is related to the peak antibody response immediately after vaccination, but protection against clinical disease or development of the carrier state lasts long after detectable antibody has disappeared. The duration of protection is unknown, present evidence indicates a period of at least 10-15 years. [Coursaget, 1994] In 1992, the World Health Organisation has recommended that universal hepatitis B vaccination should be integrated into national immunisation programmes in all countries by 1997. [Van Damme, 1997] Countries of high and intermediate endemicity, with rather uniform patterns of infection in new-borns and small children, have been more apt to implement such policies than countries of low endemicity. Within Europe, some countries (e.g. Italy, France) with relatively high incidence and prevalence of hepatitis B have initiated infant, adolescent, or combined vaccination programmes. Other countries remain unconvinced that such universal vaccination campaigns are justified. In the Netherlands, prenatal screening of pregnant women has been in effect since 1989. [Grosheide, 1991] Women are tested early in pregnancy for HBsAg status. If the mother is identified as HBsAg positive, the infant is given HBIG immediately after birth, along with a series of three doses of vaccine. In 1983, the Health Council of the Netherlands issued an advice on prevention of hepatitis B, which advocated vaccination of high-risk groups. [Gezondheidsraad, 1983] In a more recent advice of the Health Council, [Gezondheidsraad, 1996] it is concluded that this policy of selective vaccination of high-risk groups has only been partially successful. The execution of the high-risk vaccination programme has been good with regard to haemodialysis patients, haemophiliacs, medical and dentistry students, moderate with regard to institutionalised persons, new-born babies of HBsAg-positive mothers, persons who experienced needle-stick accidents and health care workers in general, and unsatisfactory with regard to sexual partners of infected individuals, male homosexuals, prostitutes and intravenous drug users. The Health Council considered insufficient funding and the lack of clearly defined responsibility for the enforcement of the policy as the main reasons for the failing immunisation policy directed at high-risk groups. In his 1996 advice, the Health Council urged a more active and intensive policy to prevent spread of HBV. In addition to the high-risk groups defined in the 1983 report, some new groups were defined, such as visitors of clinics for sexually transmitted diseases and individuals with Down’s Syndrome. The Council recommended taking steps to prepare the implementation of a general immunisation programme against hepatitis B, provided that such a general immunisation programmecould be incorporated in the Netherlands’ National Immunisation Programme and that the cost of vaccine should be acceptable..

(24) page 22 of 79. . RIVM report 403505 003. 'HVLJQDQGFRQGXFWRIOLWHUDWXUHUHYLHZ. $LPRIWKHOLWHUDWXUHUHYLHZ This literature review is being done within the framework of RIVM project 403505 “Doelmatigheid van interventies in zorg en preventie” (Efficiency of interventions in prevention and care). Besides, a cost-effectiveness analysis (CEA) of several vaccination strategies to prevent spread of hepatitis B in the Netherlands is also being conducted within RIVM project 403505. Within the literature, many publications on this topic can be found. Separate analyses have been reported for many different countries and populations (risk-groups). The purpose of the current literature review is to gain insight in methods and outcomes of previously published economic evaluations of such vaccination policies. A second purpose is to gain insight into the relevance of previous research for the current Dutch discussion on the necessity of general vaccination. This merits a limitation of the review to studies that have been performed in countries where the epidemiology of hepatitis B is comparable to epidemiological patterns as found in our country. If necessary, the design and conduct of the ongoing costeffectiveness analysis may be adjusted to findings of the literature review.. 'HVLJQRIWKHOLWHUDWXUHUHYLHZ . &ULWHULDIRUVHOHFWLRQRISDSHUV. Papers for the initialphase of the literature review were selected according to the following characteristics: • Original work on economic evaluation of hepatitis B prevention strategies, including at least one general vaccination strategy targeted at a specific risk group, a region or a country • Language: Dutch /English /French/German • Available from Dutch libraries • Since it is self-evident that there is a clear relationship between cost-effectiveness of any vaccination campaign and the prevalence of the disease, the work must be done in a country where epidemiology of hepatitis B is not too different from epidemiology of hepatitis B in the Netherlands. If details on serum-prevalence were given in the report under study, a serum-prevalence of 8 % was used as the upper limit for inclusion of a study in this review..

(25) RIVM report 403505 003. •. page 23 of 77. Sufficient quality. For each study, a quality rating was completed, according to Bradley HW DO. and Sacristan HWDO. [Bradley, 1995; Sacristan, 1993] This checklist is based on widely accepted standards of economic evaluation methodology, [Drummond, 1997] but has the additional advantage of composing a numerical score for the quality of the paper. Studies with an average quality rating ≥ 3 per applicable item were selected for the current review. The quality rating form is included in Appendix 2.. We thus excluded review articles, studies concentrating on passive immunisation and/or screening only, and studies that presented insufficient data to assess the merits of the study, such as short reports and abstracts. The reason for exclusion was documented with all excluded papers. Cost-of-illness studies were excluded from this review as well. Studies that focused only on cost-effectiveness issues within a micro context (i.e. whether or not to implement vaccination in one specific hospital) were also considered irrelevant for the current review.. . 'DWDEDVHVXVHGWRLGHQWLI\UHOHYDQWSXEOLFDWLRQV. Papers for the initialphase of the literature review were selected using the following databases: 1. COCHRANE LIBRARY. This database includes published reviews and is maintained by the Cochrane Collaboration and is available on CD-ROM (1998 4th Issue). The following search strategy was used: hepatitis B. 2. CURRENT CONTENTS (All scientific editions from 1996 until January 1999). This database was used because it contains information from important journals that were known not to be indexed in MEDLINE, such as (XURSHDQ-RXUQDORI3XEOLF+HDOWK0HGLFDO &DUHand3KDUPDFR(FRQRPLFV The following search strategy was used: Hepatitis B WITH Econom* OR Cost*. 3. DARE (Database of Reviews of Effectiveness). This database is maintained by the National Health Service Centre for Reviews and Dissemination, University of York, and is accessible through Internet. The database includes published reviews. The following search strategy was used: Hepatitis B..

(26) page 24 of 79. RIVM report 403505 003. 4. HEED (Health Economic Evaluation Database) This database is maintained by the Office of Health Economics of the Department of Health and Social Security (London) and is accessible through CD-ROM. The following search strategy was used: hepatitis. 5. INAHTA DATABASE. The International Network of Agencies of Health Technology Assessment (INAHTA) maintains a database of publications by its member organisations. This database is accessible through Internet. The following search strategy was used: hepatitis B. 6. MEDLINE (from 1966 until December 1998). MEDLINE is maintained by the United States National Library of Medicine and can be accessed through the Internet. The following search strategy was used: Hepatitis B (all subheadings) WITH Costs and Cost-Analysis (all subheadings) OR Health care Economics and Organisation (all subheadings) OR Economics (all subheadings). 7. NEED (NHS ECONOMIC EVALUATION DATABASE). This database is also maintained by the NHS Centre for Reviews and Dissemination, University of York, and is accessible through Internet. The database includes standardised descriptions of published studies. The following search strategy was used: hepatitis. Furthermore, six reviews on economic aspects of hepatitis B vaccination were identified. [Jefferson, 1994; Demicheli, 1997; Jefferson, 1998, Holliday, 1994; Badia, 1997, Beutels, 1998]. These reviews have been checked to identify further economic studies.. . )XUWKHUSURFHGXUHV. In the initialphase of the literature review, the selected papers were read. For those studies with an average quality rating ≥ 3 per applicable item, two other forms were completed: •. •. “parameter estimate” form. This form was designed to enable a more precise estimation of key parameters in the Dutch CEA of HBV prevention strategies. This form only needed completion for those parameters that were also applied in the study under review. Appendix 3 shows this parameter estimate form. “study characteristics” form, to enable a systematic description of each selected study in this literature review. This form can be found in Appendix 4..

(27) RIVM report 403505 003. page 25 of 77. The reviewer wrote a short report on the study, including both a factual description of study characteristics and a more qualitative assessment of advantages and disadvantages of the study design and conduct as chosen in the study under review. These short reports are included in Appendix 5. To enable reference tracking afterwards, relevant references were marked in the paper. If those publications had not been identified before, they were read and it was assessed whether the study should be included or not..

(28) page 26 of 79. . RIVM report 403505 003. 5HVXOWV. 5HVXOWVRIVHOHFWLRQSURFHVV The search strategies as described in paragraph 3.2. provided with a total number of 1060 entries. For the majority of these entries, it was immediately clear that no further action needed to be taken because one or more exclusion criteria applied, for instance because the publication concerned a letter, editorial comment or economic evaluation of treatment of hepatitis B or screening only. After this first selection process, 84 publications survived for a more detailed assessment. These publications have all been read carefully and a quality assessment was made. The majority of the papers selected in this phase of the review could not meet with the high methodological standards that were used for the economic evaluation methodology, did not meet all inclusion criteria or did fulfil exclusion criteria. The excluded studies are listed in Appendix 6. Finally, 16 papers fulfilled all requirements and were selected for further investigation. These 16 papers are described in more detail in Appendix 5 and summarised systematically in the following paragraphs.. 0DLQFKDUDFWHULVWLFVRIVHOHFWHGSDSHUV . 9DFFLQDWLRQVWUDWHJ\. Table 2 shows the different vaccination strategies and target populations that are described in 16 selected papers..

(29) RIVM report 403505 003. page 27 of 77. Table 2: Vaccination strategies and target populations as studied in 16 selected papers First author Mulley. Year of publication 1982. Vaccination strategy. Target population Homosexual men Surgical residents General US population All neonates. Ginsberg. 1992a. Ginsberg. 1992b. General vaccination Screen + vaccinate Passive immunisation General vaccination Doing nothing General vaccination. Demicheli. 1992. General vaccination. Bloom. 1993. General vaccination Screen + vaccinate Doing nothing. Krahn. 1993. Oddone. 1993. Eelkman Rooda. 1994. Margolis. 1995. Antoñanzas. 1995. Kerleau. 1995. Screen + vaccinate Screen + vaccinate + universal vaccination of new-borns Doing nothing General vaccination Doing nothing Screen + vaccinate General vaccination Doing nothing Screen + vaccinate Screen + vaccinate + universal vaccination of new-borns Screen + vaccinate + universal vaccination at 11-13 years of age General vaccination Screen + vaccinate + general vaccination General vaccination Screen + vaccinate Doing nothing. Fenn. 1996. General vaccination. All under-16-year-olds All neonates All neonates + 12 years ‘catchup’ of 12 year-olds All neonates 10-year-olds High-risk adults General adult population Mixes of these 4 populations All neonates. Pre-dialysis patients Patients starting dialysis All neonates All neonates + 10 years ‘catchup’ of 10 year-olds All neonates 11-13-year-olds. 12-year-olds neonates Combinations of 2 groups General population 15-40 year-old males Male homosexuals Drug users Neonates 6-year-olds 11-year-olds Neonates + 11-year-olds.

(30) page 28 of 79. First author. RIVM report 403505 003. Beutels. Year of publication 1996. Vaccination strategy. Target population. General vaccination Doing nothing. All neonates 12-year-olds All neonates + 12 years ‘catchup’ of 12 year-olds All neonates 12-13-year-olds All neonates + ‘catch-up’ of 1213-year-olds All neonates 10-year-olds 12-year-olds 12-year-olds Neonates. Garuz. 1997. Screen + vaccinate General vaccination. Wiebe. 1997. Screen + vaccinate + general vaccination. Krahn. 1998. General vaccination Screen + vaccinate + high-risk individuals. Table 2 shows that the majority (13 out of 16) of selected studies at least describe the results of neonatal vaccination strategies. Several studies combine this strategy with the strategy of screening of pregnant women and selective vaccination of their offspring, because this was the major preventive strategy in a country, before general vaccination was considered. Eleven studies describe results of adolescent vaccination campaigns. Only three studies were directed at specific risk-groups, such as drug users or male homosexuals.. &RXQWU\ RI VWXG\ DQG LWV HSLGHPLRORJLFDO SDWWHUQ RI +%9 VSUHDG Table 3 provides information on the country or region where the study was performed, along with its epidemiological pattern of HBV transmission, according to the authors of the papers. The incidence data are in general calculated data, based on more or less sound assumptions on the percentage of underreporting of hepatitis B..

(31) RIVM report 403505 003. page 29 of 77. Table 3: Country of study and epidemiological data First author Mulley Ginsberg Ginsberg Demicheli Bloom Krahn. Year of publication 1982 1992a 1992b 1992 1993 1993. Oddone Eelkman Rooda Margolis Antoñanzas Kerleau Fenn Beutels Garuz. 1993 1994 1995 1995 1995 1996 1996 1997. Wiebe Krahn. 1997 1998. Country, region United States Israel Israel Italy United States Canada & United States United States Netherlands United States Catalonia, Spain France United Kingdom Flanders, Belgium Spain Manitoba, Canada British Columbia, Canada & USA. a. not given. b. risk-group specific incidence. c. depends on age-group, no overall figure given. Incidence / 100.000 population 100 270 270 30 115 a (300.000 infections per year) 200 b. Prevalence of markers 5 – 10 %. c. a. a. 4.8 %. 150 100 12.6 20 – 500 c a (60.000 infections per year) 19 30 (British Columbia). a. a a a. 4.8 % a. a. 5 – 10 % a. 7.6 % a. a a. It appears from table 3 that, despite the DSULRULselection of studies from countries with low or intermediate endemicity, large differences in reported or calculated incidence rates appear between the studies. Some studies do not report on the epidemiological background at all. Only very few studies have been done in countries where the epidemiological pattern of hepatitis B is really comparable with the Dutch epidemiological pattern. This hampers the direct comparison of study results between the selected studies.. . 9DULDEOHVFRQVLGHULQJWKHYDFFLQDWLRQSURJUDP. Table 4 shows the different assumptions that have been made on: 1. the duration of immunity after successful vaccination; 2. the efficacy of the vaccine (after a completed cycle of 3 vaccinations); 3. the compliance with the vaccination program; 4. the percentage of the target population that is successfully vaccinated by the vaccination program; 5. whether a booster vaccination was assumed and if so, after how many years; 6. whether adverse effects of the vaccine were included in the study..

(32) page 30 of 79. RIVM report 403505 003. Table 4: Assumptions on effectiveness of vaccination program (for explanation of numbers see above) First author Mulley Ginsberg Ginsberg Demicheli Bloom Krahn. Year of publication 1982 1992a 1992b 1992 1993 1993. Oddone Eelkman Rooda Margolis Antoñanzas Kerleau Fenn Beutels Garuz. 1993 1994 1995 1995 1995 1996 1996 1997. Wiebe. 1997. Krahn. 1998. 1.. 2.. 3.. 4.. 5.. 6.. 5 years 5 years a 5 years a life-time life-time decreasing annually c 3 years life-time 6-13 years life-time 5 years 5 years 35 years 10 years + life-time decreasing annually c decreasing annually e. 87.5 % 95 % 95 % 100 % 80-90 % b 90 %. 100 % 100 % 95 % 100 % 33-70 % b 90 %. 87.5 % 95 % 90.25 % 100 % 26-63 % b 81 %. 10 + 20 -. yes yes yes no no yes. 55-80 % d 100 % 95-100 % 90-95 % 95 % 90 % 95-99 % 90 %. 100 % 100 % 40-60 % 68-90 % 100 % 100 % 80-90 % 68-90 %. 55-80 % 100 % 69 % 61-86 % 95 % 90 % 76-89 % 61-81 %. 10 5 10. no no no no no no no no. 90 %. 90 %. 81 %. -. yes. 99 %. 94 %. 93 %. -. no. a. efficacy of vaccine declining by 15 % every 5 years, so efficacy is 80.75 % after 10 years, 68.64 % after 15 years, et cetera. b. depending on target population. c. by 2-8 %. d. lower response rate because of diseased target group. e. by 4.5 %. Table 4 shows convincingly that the 16 studies have made quite different assumptions on the effectiveness of the vaccination programs. For instance, evidence on the duration of the protective effect of vaccination is still lacking, because the recombinant vaccine was only introduced in 1986. With respect to the duration of immunity after successful vaccination, some studies make a very conservative assumption of 3-5 years protection, while others assume lifetime immunity. Several studies tried to handle the uncertainty by way of modelling annually decreasing immunity levels. The fourth column of table 4 shows the overall effectiveness of the program, which is a product of assumptions on vaccine efficacy and compliance with the program. Depending on the target population, percentages as low as 26 % and as high as 100 % have been found..

(33) RIVM report 403505 003. . page 31 of 77. 7\SHRIHFRQRPLFHYDOXDWLRQDQGSHUVSHFWLYHRIVWXG\. There are three major types of economic evaluation that consider both costs and effects of a technology. These are cost-effectiveness analysis (CEA), cost-utility analysis (CUA) and cost-benefit analysis (CBA). • •. •. In a CEA, competing interventions are compared in terms of cost per unit of consequence, usually life-years gained or cases of HBV infection averted. A CUA design enables the measurement of effects on both quantity and quality of life. Competing interventions are compared in terms of cost per unit of utility gained, usually cost per Quality-Adjusted-Life-Year (QALY). In a CBA, all costs and benefits of health care programs are measured in monetary units. Hence, a CBA requires monetary valuation of quality of life outcomes, which is a rather difficult task [Drummond, 1997]. Some studies that omit the effects on quality of life are labelled as a CBA, but are in fact CEAs. The two selected Ginsberg studies are good examples for this type of mislabelling.. In table 5 an overview of the economic evaluation study design in the selected studies is given, along with the perspective that was taken for the evaluation. The application of strict selection criteria for our review did not allow for a study to take a micro perspective only, such as economic effects for one hospital. Therefore, two major perspectives remain: the societal perspective and the third-party-payer perspective (including a national health services perspective)..

(34) page 32 of 79. RIVM report 403505 003. Table 5: Type of economic evaluation and perspective of the evaluation First author Mulley Ginsberg. Year of publication 1982 1992a. Type of economic evaluation CEA CEA. Ginsberg. 1992b. CEA. Demicheli Bloom Krahn. 1992 1993 1993. CBA CEA CEA. Oddone Eelkman Rooda Margolis Antoñanzas Kerleau Fenn Beutels. 1993 1994 1995 1995 1995 1996 1996. CEA CEA CEA CEA CEA CEA CEA. Garuz Wiebe Krahn. 1997 1997 1998. CEA CEA CEA. Perspective taken Third-party payer Societal Third-party payer Societal Third-party payer Societal Third-party payer Societal Third-party payer Third-party payer Societal Societal Third-party payer Third-party payer Third-party payer Societal Third-party payer Third-party payer Third-party payer Societal Third-party payer. Table 5 makes clear that CEA is the study design that is applied most often. As mentioned before, the two Ginsberg studies represent also CEAs despite the fact that their authors call them CBAs. Only one CBA and no CUA were selected. The majority of the studies at least reported results from a third-party payer perspective, either a health insurance or a national health services system. Half of the studies also took a societal perspective and tried to incorporate all costs and effects of interventions on a societal level, irrespective of the paying or benefiting institution..

(35) RIVM report 403505 003. . page 33 of 77. 0RGHOOLQJWHFKQLTXHV. The different studies have used different techniques to model the effectiveness of the intervention under survey. Table 6 provides with an overview of this aspect of the selected studies. Table 6: Modelling technique(s) used in selected studies First author Mulley Ginsberg Ginsberg Demicheli Bloom Krahn Oddone Eelkman Rooda Margolis Antoñanzas Kerleau Fenn. Year of publication 1982 1992a 1992b 1992 1993 1993 1993 1994 1995 1995 1995 1996. Beutels Garuz. 1996 1997. Wiebe Krahn. 1997 1998. Modelling technique(s) Decision analysis Decision analysis Decision analysis “Brown’s exponential smoothing” forecasting technique Decision analysis Decision analysis + Markov chain analysis Decision analysis Markov chain analysis Decision analysis Markov chain analysis + “Compartmental dynamic model” Simple calculations of averted infections over 5 year period Markov chain analysis with accommodation for “herd immunity effect” by means of changing transition probabilities Decision analysis Mathematical model based on differential equations, Markov chain analysis + Monte Carlo simulation Decision analysis + Markov chain analysis Markov chain analysis. It appears from table 6 that decision analysis is the modelling technique that is used most often. Most studies that used this technique applied the cyclic algorithm designed by Mulley et al. [Mulley, 1982] and adapted it to fit the needs of the specific study. However, in recent years, more complex techniques such as Markov chain analysis and more mathematically oriented modelling techniques have been applied more often. Only few studies tried to accommodate for indirect immunisation effects, the so-called “herd immunity” effect (the reduction of infections over time due to the smaller stock of chronic carriers as the vaccination program proceeds over time). The use of static modelling techniques for the estimation of the effects of vaccination programs in general leads to overestimation of the benefits of these programs.. 'XUDWLRQRIYDFFLQDWLRQSURJUDPVDQGWLPHKRUL]RQIRUFDOFXOD WLRQRIFRVWVDQGHIIHFWV Table 7 shows the differences between selected studies in duration of vaccination programs on the one hand and the number of years for which the effects and costs of these programs were calculated on the other hand..

(36) page 34 of 79. RIVM report 403505 003. Table 7: Duration of health care intervention and time horizon used to calculate effects and costs First author Mulley Ginsberg Ginsberg Demicheli Bloom Krahn Oddone Eelkman Rooda Margolis Antoñanzas Kerleau Fenn Beutels Garuz Wiebe Krahn. Year of publication 1982 1992a 1992b 1992 1993 1993 1993 1994 1995 1995 1995 1996 1996 1997 1997 1998. Duration of vaccination program 1 year 45 years 36 years 99 years 1 year 1 year 1 cohort of dialysis patients 100 years 1 year 12 years 5 years 1-25 years 1 year 1 year 1 year 1 year. Time period effects. Time period costs. 1 year 110 years 70 years 99 years 30 years life-time 3 years. 20 years 110 years 70 years 129 years 60 years life-time 10-30 years. 100 years 45 years 22-32 years 5 years 110 years 35 years 10-30 years life-time life-time. 100 years 45 years 22-32 years 25 years 110 years 70 years 10-30 years life-time life-time. Table 7 shows that the different studies have used quite different time periods for both the vaccination program itself and the calculation of the effects and costs over time. The duration of the program varied between 1 and 100 years. The time horizon for the calculation of the effects of programs varied between 1 and 110 years. The maximum time horizon for the calculation of costs was even longer, at 129 years..

(37) RIVM report 403505 003. . page 35 of 77. &RVWVRIYDFFLQH. Table 8 provides with information on the parameters that were used for the costs of vaccine (for a complete cycle of 3 vaccinations), the source that was mentioned for the costs of the vaccine and the year of costing that was used for all cost data in the study. Table 8: Costs of vaccine (currency), source that was used for vaccine costs and base year for all other costing First author Mulley Ginsberg Ginsberg Demicheli Bloom. Year of publication 1982 1992a 1992b 1992 1993. Krahn Oddone Eelkman Rooda Margolis. 1993 1993 1994 1995. Antoñanzas. 1995. Kerleau Fenn Beutels Garuz. 1995 1996 1996 1997. Wiebe. 1997. Krahn. 1998. a. Costs of vaccine and currency US$ 100 a US$ 6.19 US$ 6.57 US$ 10 US$ 160 (new-borns) US$ 225 (adolescents) Can$ 39.30 US$ 114. Source Charges Manufacturer Manufacturer Literature Hospital pharmacy Not given Hospital pharmacy. b. b. US$ 20 (new-borns) US$ 40 (adolescents) PTA 2400 (new-borns) PTA 3900 (adolescents) FF 420 £ 26 BEF 1200 US$ 21.2 (new-borns) US$ 28.4 (adolescents) Can$ 14 (new-borns) Can$ 28 (adolescents) US$ 20. “Federal contract price” Manufacturer. Base year for costing data 1980 1990 1990 Not given 1989 1990 1987 1991 1993 1992. Charges Not given Manufacturer Manufacturer. 1991 1992/1993 1995 1993. Not given. 1993. Ministry of Health contract price. 1994. including fees for handling and administration. b. the cost of vaccine was not included in the Base Case Scenario, because study was set up as a threshold analysis (at what vaccine price can a vaccination campaign be organised in a cost-neutral way?). Table 8 shows that large differences in the assumptions on the costs of vaccine appear within the selected studies. The lowest vaccine costs that were found were as low as US $ 6 (1989 cost data) for a complete cycle, the highest costs ranged to US $ 160 per cycle of 3 doses (1990 cost data)..

(38) page 36 of 79. . RIVM report 403505 003. 'LIIHUHQWFRVWGDWDLQFOXGHGLQVWXGLHV. Three major types of cost data may be distinguished in economic evaluations: direct, indirect and intangible costs. Direct costs consist of both the costs for the vaccination program and the costs for the treatment of hepatitis B and its long-term sequelae (e.g. acute hepatitis, cirrhosis, and hepatocellular carcinoma). Indirect costs are generated by productivity loss due to morbidity or premature mortality. Intangible costs are the costs associated with suffering and pain due to ill health, and, in case of premature death, the value of human life itself. Table 9 provides with information on the different cost categories that were included in the selected studies. Table 9: Different types of cost data included in the selected studies First author Mulley Ginsberg Ginsberg Demicheli Bloom Krahn Oddone Eelkman Rooda Margolis Antoñanzas Kerleau Fenn Beutels Garuz Wiebe Krahn. Year of publication 1982 1992a 1992b 1992 1993 1993 1993 1994 1995 1995 1995 1996 1996 1997 1997 1998. Direct costs yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes yes. Indirect costs no yes yes yes no yes no yes yes no no no yes no no yes. Intangible costs no no no yes no no no no no no no no no no no no. Not surprisingly, all studies incorporated direct medical costs. Indirect costs were considered in 50 percent of the selected studies. Only one study attempted to calculate intangible costs. About half of the studies reported an in-depth study of direct medical costs. Such in-depth cost studies consist for instance of reviews of patient charts to gain insight in levels of medical consumption in different stages of disease, expert panels on the most likely course of a chronic hepatitis B infection, and detailed studies on actual costs of health care, not charges. The other half of the studies either quoted data from the more sophisticated studies, or simply made estimates and assumptions from the literature or consulted one or two experts. Those studies that included indirect costs all made assumptions on the number of working days lost due to hepatitis B..

(39) RIVM report 403505 003. . page 37 of 77. 'LVFRXQWLQJ. Because any intervention directed at the prevention of hepatitis B infections requires that investments have to be made at present, while benefits of this investment are also realised in the distinct future, it is essential that a correction for these differences in time perspective is made. Health economists do agree on the fact that future costs need to be discounted, that means expressed in today’s currency. Therefore, future costs weigh less heavily than today’s costs. The annual discount rate may differ between countries and is also time-pendent. Relevant recommendations for this rate are: • • •. 3% (Panel on Cost-Effectiveness in Health and Medicine [Gold, 1996]) 4% (Dutch Guidelines for Pharmaco-Economic Research [CVZ, 1999] and Dutch Ministry of Finance [Ministry of Finance, 1996]) 5% (Canadian Guidelines for Economic Evaluation of Pharmaceuticals [CCOHTA, 1997]). However, there is ongoing controversy among health-economists on the necessity to discount future benefits as well. [van Hout, 1998] Table 10 shows the policies that selected studies have chosen with regard to discounting of future costs and benefits, and the discount rate that was used in the Base Case calculations. Table 10: Discounting of future costs and benefits and discount rate used First author Mulley Ginsberg Ginsberg Demicheli Bloom Krahn Oddone Eelkman Rooda Margolis Antoñanzas Kerleau Fenn Beutels Garuz Wiebe Krahn a. Year of publication 1982 1992a 1992b 1992 1993 1993 1993 1994 1995 1995 1995 1996 1996 1997 1997 1998. Discounting of costs yes yes yes yes yes yes yes no a yes yes yes yes yes yes yes yes. Discounting of benefits no no no no no/yes yes no no a no/yes no/yes no no/yes no/yes no/yes yes no/yes. Discount rate 6% 7.5 % 7.5 % 8% 5% 5% 5% 0%a 5% 5% 8% 6% 5% 5% 5% 3%. Baseline results presented undiscounted, but 5 and 10 % discount rates applied in sensitivity analyses.

(40) page 38 of 79. RIVM report 403505 003. It appears from table 10 that only one study chose to present undiscounted cost data. Discount rates varied between 3 and 8 percent in the baseline calculations of results. By the choice of a more conservative (lower) discount rate, more weight is attached to future costs and benefits than by the choice of a higher discount rate. About half of the studies resolved the controversy on the necessity to discount future benefits by presenting both discounted and undiscounted data on the benefits of the vaccination program. Almost all studies evaluated the influence of the choice for a specific discount rate in sensitivity analyses (see further).. 6HQVLWLYLW\DQDO\VLV One of the elements of a good economic evaluation is that the influence of the assumptions made and the parameter values chosen are subjected to a sensitivity analysis. The aim of a sensitivity analysis is to study whether conclusions that are drawn in the baseline analysis would change if other baseline parameter values were chosen. Table 11 shows which parameters were subjected to sensitivity analyses in the different studies, and which parameters appeared to be most influential for the final results of the study..

(41) RIVM report 403505 003. page 39 of 77. Table 11: Variables included in the sensitivity analyses and most influential variables First author Mulley. Year of publication 1982. Ginsberg. 1992a. Ginsberg. 1992b. Demicheli. 1992. Bloom. 1993. Krahn. 1993. Oddone. 1993. Eelkman Rooda. 1994. Margolis. 1995. Variables included. Most influential variables. Costs of HBV infection, Vaccine efficacy, Vaccination costs, Natural history of HBV infection, HBV incidence in target population, The inclusion of indirect costs Discount rate, Vaccine effectiveness decay, Degree of underreporting of acute hepatitis B cases, The inclusion of indirect costs Discount rate, Vaccine effectiveness decay, Degree of underreporting of acute hepatitis B cases, Compliance with vaccination, Vaccination costs, The inclusion of indirect costs HBV incidence, Discount rate, Costs of HBV infection, The inclusion of indirect and intangible costs. Screening costs, Vaccination costs, Costs of HBV infection, HBV incidence, The inclusion of indirect costs. Vaccination costs, Vaccine efficacy, Natural history of HBV infection, Discounting of non-monetary benefits Vaccination costs, Discount rate, Prevalence of carrier state, HBV incidence, Percentage of infections progressing to chronic stage, Natural history of HBV infection, Duration of vaccine efficacy, Compliance with vaccination, Costs of HBV infection Vaccine efficacy, Costs of HBV infection, Vaccination costs, HBV incidence HBV incidence, Natural history of HBV infection, Inclusion of costs of screening blood donors, Costs of HBV infection, Vaccine efficacy, Discount rate, The inclusion of indirect costs Vaccination costs, Compliance with vaccination, Costs of HBV infection, Discount rate, The inclusion of indirect costs, Percentage of infections progressing to chronic stage, Natural history of HBV infection, Prevalence of. Discount rate, The inclusion of indirect costs. Vaccination costs, Discount rate, The inclusion of indirect costs. HBV incidence, Discount rate, Costs of HBV infection, The inclusion of indirect and intangible costs Vaccination costs, Discounting of non-monetary benefits Vaccination costs, Discount rate, Percentage of infections progressing to chronic stage, HBV incidence, Du-ration of vaccine efficacy. Vaccination costs, HBV incidence, Vaccine efficacy Inclusion of costs of screening blood donors, Discount rate, The inclusion of indirect costs Discount rate, Vaccination costs, Prevalence of HBV infection (estimated life-time risk of infection), (influence of other variables not reported).

(42) page 40 of 79. First author. RIVM report 403505 003. Year of publication. Antoñanzas. 1995. Kerleau. 1995. Fenn. 1996. Beutels. 1996. Garuz. 1997. Wiebe. 1997. Krahn. 1998. Variables included HBV infection (estimated life-time risk of infection) Transplant probability, Years of followup, Cirrhosis costs, Compliance with vaccination, Vaccination costs, Discount rate, Costs of HBV infection, Percentage of infections progressing to chronic stage, HBV incidence Discount rate, Compliance with vaccination, Vaccination costs, Costs of HBV infection, HBV incidence HBV incidence, Degree of underreporting of acute hepatitis B cases, Percentage of infections progressing to chronic stage, Risk of cirrhosis, Risk of HCC, Costs of HBV infection, Organisational costs, Discounting of nonmonetary benefits Discount rate, Natural history of HBV infection, The inclusion of indirect costs, Vaccination costs Costs of HBV infection, Discount rate, Vaccination costs, Epidemiological assumptions (not specified), Time horizon for calculation of effectiveness, Increase of indirect costs Discount rate, Vaccine effectiveness decay, HBV incidence, Vaccination costs, Costs of HBV infection, Compliance with vaccination, Natural history of HBV infection HBV incidence, Vaccination costs, Degree of underreporting of acute hepatitis B cases, Percentage of infections progressing to chronic stage, Discount rate, Vaccination costs, Duration of vaccine efficacy, Costs of HBV infection. Most influential variables. Vaccination costs, Discount rate, Percentage of infections progressing to chronic stage, HBV incidence, Discount rate HBV incidence, Vaccination costs Discounting of nonmonetary benefits, HBV incidence, Degree of underreporting of acute hepatitis B cases, Percentage of infections progressing to chronic stage, Risk of cirrhosis Discount rate, The inclusion of indirect costs, Vaccination costs Vaccination costs, Time horizon for calculation of effectiveness. Discount rate, Vaccine effectiveness decay, HBV incidence, Vaccination costs. HBV incidence, Vaccination costs (influence of other variables not reported). In reporting the results of sensitivity analyses, not all studies have applied similar reporting standards. Therefore, it was sometimes difficult to assess the outcomes of sensitivity analyses. Some authors have used relative changes towards the baseline estimates as a measure of robustness of results. In these cases, it was relatively easy to assess and control the results of.

(43) RIVM report 403505 003. page 41 of 77. sensitivity analyses. However, other studies have used some arbitrary threshold of costeffectiveness, such as cost per life year gained not exceeding US $ 50.000. As long as changing baseline assumptions did not result in cost-effectiveness ratios above this arbitrary threshold, it was concluded that outcomes were insensitive to changes in baseline assumptions. Such conclusions are much more difficult to control for readers of these papers. The data in the last column of Table 11 should therefore not be regarded as an exhaustive overview of all sensitive parameters, but merely as a minimum set of parameters that appear important for the determination of the robustness of the models used. The following parameters were most often shown to have a significant impact on the results of a study: 1. 2. 3. 4. 5. 6. 7.. Vaccination costs Discount rate HBV incidence The inclusion of indirect costs/benefits The percentage of infections progressing to chronic stages of disease Discounting of non-monetary benefits Costs of HBV infection, including long-term consequences such as cirrhosis and liver cancer. It is also possible to create a shortlist of parameters that in general seem to have a rather limited influence on the cost-effectiveness of HBV vaccination programmes: • The natural history of chronic HBV infection (the percentage of patients that will develop long-term sequelae of HBV infection). This excludes the percentage of patients progressing to the chronic stage of disease, as mentioned above. • Factors associated with vaccine effectiveness: the efficacy of the vaccine, compliance with vaccination, the duration of vaccine efficacy, and the decay in vaccine efficacy over time.. &RQFOXVLRQV UHDFKHG RQ HFRQRPLF DWWUDFWLYHQHVV RI HYDOXDWHG SURJUDPV The cost-effectiveness and cost-benefit ratios that were reported in the 16 papers are shown in Table 12. These ratios refer to the baseline outcomes of studies. If possible, the range in outcomes is provided as well..

(44) page 42 of 79. RIVM report 403505 003. Table 12: Cost-effectiveness and cost-benefit ratios (with ranges) as reported in baseline outcomes of selected studies (per program evaluated) First author Mulley. Year of publication 1982. Cost-effectiveness ratio. Benefit-cost ratio. Ranges. •. Cost per case prevented: US$ 10.000 (attack rate 0.3 %) US$ 2.000 (attack rate 1 %) US$ 0 (attack rate 5 %). Benefit-cost ratio > 1 in populations with annual attack rates above 5 %. -. •. Ginsberg. 1992a. -. Benefit-cost ratios of neonatal vaccination program: 1.56 (direct cost only) 2.80 (direct and indirect cost). In sensitivity analyses: 0.68-21.31 (direct costs only) 1.23-27.01 (direct + indirect costs). Ginsberg. 1992b. -. Benefit-cost ratios of under16-year-olds vaccination program: 1.88 (direct cost only) 3.57 (direct and indirect cost). In sensitivity analyses: 0.80-42.10 (direct costs only) 1.48-122.97 (direct + indirect costs). Demicheli. 1992. -. Benefit-cost ratio of neonatal vaccination program always < 1 in basecase. Benefit-cost ratio >1 if high future costs of HBV infection were assumed (US$ 17 million! per case). Remarks and conclusions. Imprecise data because abstracted from figure • annual attack rate in general population = 0.1 % • General vaccination is probably cost saving in I.V. drug users and homosexual men • Benefit to cost ratio only < 1 if discount rate of 10 % used • Neonatal vaccination strategy is cost saving for both health care system and society • Benefit to cost ratio only < 1 if discount rate of 10 % used • Under-16-year-olds vaccination strategy is cost saving for both health care system and society • With epidemiological pattern of declining incidence of HBV infections, general vaccination will most likely never be an efficient strategy.

(45) RIVM report 403505 003. First author Bloom. Krahn. page 43 of 77. Year of publication 1993. 1993. Cost-effectiveness ratio. Benefit-cost ratio. Ranges. •. Cost per life-year saved: US$ 375-3.695 (mixed childhood strategy) US$ 3.066-38.632 (newborns) US$ 13.938-97.256 (adolescents) US$ 54.524-257.418 (adults) Cost per life-year saved: (newborn strategy in comparison with selective strategy): US$ 30.347 (societal perspective) US$ 57.197 (third party payer perspective). Oddone. 1993. Cost per life-year saved: US$ 583.333 (predialysis) US$ 493.902 (dialysis patients). Eelkman Rooda. 1994. 6WXG\UHVXOWVQRWUHSRUWHGLQ VWDQGDUGZD\. •. •. • From a societal perspective: dominant (both lower cost and longer life-expectancy) to US$ 214.000 From 3rd party-payer perspective: US$ 11.000-US$ 253.000 Cost per life-year saved < US$ 50.000 at annual attack rates > 2 %. • •. •. Remarks and conclusions lower limit referring to undiscounted benefits, higher limit referring to discounted benefits Universal vaccination of newborns is cost-saving at vaccine price below US$ 34 Mixed childhood vaccination strategy is most cost-effective Cost-effectiveness ratios all refer to discounted benefits Vaccination of newborns is cost-saving at vaccine price below US$ 7. Cost-effectiveness ratios all refer to undiscounted benefits • local incidence rates important to choose strategy • Vaccination of newborns + catch-up strategy is cost-saving at vaccination price of NLG 68 (5 % discounted costs).

No results found